Garret N.

Vanderplaats

Structural Optimization for Statics,

Garret N. Vanderplaats
Vanderplaats Research & Development, Inc
1767 S. 8th Street
Colorado Springs, CO 80906 USA
vanderplaats@vrand.com
Dynamics and Beyond
Structural optimization has matured to the point that it can be routinely applied to a wide
range of real design tasks. The purpose here is threefold. First, the general optimization
task will be defined. Second, the state of the art in structural optimization will be reviewed.
Finally, examples will be presented to demonstrate the level of sophistication possible in
applying this technology. It is concluded that, while much research always remains,
optimization technology has matured to the point where it can and should be used
routinely for engineering design.
Keywords: Optimization, structural optimization, automated design

Introduction
With his landmark paper in 1960, Schmit ushered in over forty
years of intensive development in structural and general purpose
optimization research. This has culminated in numerous commercial
products that are available today to solve design problems of
remarkable size and complexity. These basic developments, together
with modern graphical interfaces, makes it possible to use this
technology with very little formal training in optimization theory.1
Despite the widespread availability of this technology, it is
seldom taught as a design tool by universities and remarkably
underutilized by industry. Yet the motivation to use optimization is
compelling. For automobiles, a ten percent mass reduction will
increase fuel economy by about seven percent. Only a one percent
economy improvement will save nearly three billion dollars per year
in the U.S. at the pump. Similarly, by reducing the mass of a
commercial aircraft by about two hundred pounds adds a paying
passenger for the life of the aircraft. A one pound reduction in the
mass of a spacecraft will either add a pound of payload or save
about $20,000 per flight to space. The list of examples could
continue for pages. Even beyond the cost argument, the savings in
natural resources through the use of optimization could be immense.
The purpose here is to briefly review the development of
structural optimization leading to the current state of the art
and offer examples to demonstrate the power of optimization
to enhance the design process
Nomenclature
F = force in member
F(X) = objective function
gj(X) =j-th inequality constraint
hk(X) = k-th equality constraint
K = stiffness matrix
l = number of equality constraints
L = length of member
m = number of inequality constraints
n = number of design variables
P = structural load vector
S = search direction
U = vector of structural displacements
X = vector of design variables
X = single design varible
δX = change in design variables
Greek Symbols
Presented at XI DINAME – International Symposium on Dynamic Problems of
Mechanics, February 28th - March 4th, 2005, Ouro Preto. MG. Brazil.
Paper accepted: June, 2005. Technical Editor: Joé Roberto de França Arruda.
α = move parameter
σ = stress
σ = allowable stress
σijk = stress in element i, component j, load case k
∂ = partial derivative
∇ = gradient operator
Subscripts
i design variable number
j inequality constraint number
x derivative with respect to x
Superscripts
L lower bound on design variable
New new design
Old old design
U upper bound on design variable
-1 inverse
What is Design Optimization?
Optimization is intrinsically tied to our desire to excel, whether
we are an athlete, artist or engineer. We all adjust some parameters,
perhaps our time, to minimize or maximize one or more results such
as income, leisure time or job satisfaction. We do this subject to
limitations or constraints, such as physical ability, time available,
legal restrictions or moral codes of conduct. Thus, whatever our
field of endeavor, we constantly strive to solve a constrained
optimization problem.
In engineering, we create products. To do this, we normally use
computer analysis to judge the quality of our designs. We use
computational fluid dynamics codes to calculate energy
requirements and flow patterns in a ducting system. We use finite
element analysis to calculate stresses, deflections, vibration
frequencies, etc. of a structure. In almost all disciplines, we use
computational, and sometimes experimental, tools to judge the
quality of our proposed designs. If not satisfactory, we modify the
design and perform repeated analyses in an effort to improve the
product, or at least meet the design requirements.
This traditional approach of analyze and revise normally
involves only changing a few variables (often only one) at a time
and does not account very well for the interaction among the
variables.
Now imagine we can change large numbers of design
parameters simultaneously in order to improve the design while
satisfying all design requirements, at the same time accounting for
the interactions among the parameters. This is exactly what
numerical optimization does.
Our computer analysis program has a set of inputs that we may
consider to be contained in a vector X. Based on this input, the
analysis provides outputs. One or more of these outputs can be

316 / Vol. XXVIII, No. 3, July-September 2006 ABCM